CN114282816A - Give birth to bright district large screen electronic tags automatic management system - Google Patents
Give birth to bright district large screen electronic tags automatic management system Download PDFInfo
- Publication number
- CN114282816A CN114282816A CN202111607248.0A CN202111607248A CN114282816A CN 114282816 A CN114282816 A CN 114282816A CN 202111607248 A CN202111607248 A CN 202111607248A CN 114282816 A CN114282816 A CN 114282816A
- Authority
- CN
- China
- Prior art keywords
- goods
- commodity
- area
- unit
- fresh
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention discloses an automatic management system for a large-screen electronic tag of a fresh area, which comprises a base station, a first shooting device, an area management and control device and a remote server, wherein the base station is used for shooting a fresh area; the region control device comprises a price change data table, a price adjusting module, a trend analyzing module and a plurality of freshness evaluation models of fresh commodities; the price change data table is used for storing the types of the fresh commodities, the freshness grades of the commodities and corresponding price values; the freshness evaluation model evaluates the overall freshness grade of the specified fresh commodity and feeds back the evaluation result to the trend analysis module; and the trend analysis module calculates the freshness change trend of the fresh-keeping object, and generates a replenishment instruction or a price adjustment instruction according to the freshness change trend. The invention greatly reduces the time for manually adjusting and confirming the commodity information, has high reliability of the verification result, ensures that the commodity type and the price label which can be observed by the user are displayed consistently, and improves the user experience.
Description
Technical Field
The invention relates to the technical field of electronic tags in fresh food areas, in particular to an automatic management system for a large-screen electronic tag in a fresh food area.
Background
The electronic tag using the electronic paper has advantages of excellent display effect, convenience in price updating and the like, is popularized in supermarkets, and gradually replaces the original paper tag. When the commodity price needs to be adjusted, the price updating within the whole commodity exceeding range can be realized only by sending the adjusted commodity price image to the corresponding electronic tag monomer. The electronic tags are various in size, for example, in a supermarket, the electronic tags with the size of 4-6 inches can be selected on shelves in various department goods areas, the electronic tags with the size of more than 10 inches can be selected in a fresh goods area, and price prompt service is provided for customers in a mode that the electronic tags are hung above fresh goods.
At present, most of electronic tags are updated through manual background processing, and a few research schemes propose automatic updating of electronic tags after commodities are identified through some reading equipment. The invention with application number 201810299344.5, which is published in 2019, 10, 18 and discloses an automatic commodity information identification method, which can be used for automatically adjusting commodity information on an electronic shelf label by combining with an automatic commodity identification technology, greatly improves the working efficiency and simultaneously avoids information inconsistency caused by commodity misdistribution. However, in the application, the RFID reader is used to identify the information of the goods, and only the electronic tag corresponding to a single goods can be updated, and once the electronic tag corresponds to a plurality of goods and the position area occupied by the goods changes, the price tag cannot be automatically updated. At present, a price label automatic updating method for a large-screen electronic label for simultaneously displaying a plurality of commodities does not appear.
In addition, the quality guarantee period of the fresh commodities is short, the selling probability of a plurality of fresh commodities can be reduced due to reasons such as water loss in the same day, and the selling probability can be improved by adopting modes such as price reduction and sales promotion by common merchants. At present, for the phenomenon, workers change commodities in a fresh area in the background in a manual mode after checking the current situation of the commodities, and then display the changed results on an electronic tag.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides the large-screen electronic tag automatic management system for the fresh area, which can estimate the freshness grade of the commodities in the fresh area, provide a solution for replenishment or price reduction promotion according to the estimation result, automatically adjust the commodity price tags according to the promotion scheme preset by a user, and greatly reduce the time for manually adjusting and confirming the commodity information.
In order to achieve the purpose, the invention adopts the following technical scheme:
an automatic management system for a large-screen electronic tag in a fresh area is used for managing and updating display contents of electronic tags distributed and installed in the fresh area; the automatic management system comprises a base station, a first shooting device, an area management and control device and a remote server;
the remote server locally generates a price tag of each fresh area commodity and sends the price tag to a corresponding electronic tag through the base station; the electronic tag adopts an electronic paper display screen and is used for displaying the price tag of at least one fresh area commodity on the shelf;
the first shooting device is arranged below the electronic tag and connected with the area control device, shoots fresh commodity images of the specified goods placing area corresponding to the electronic tag and sends shooting results to the area control device;
the region control device comprises a price change data table, a price adjusting module, a trend analyzing module and a plurality of freshness evaluation models of fresh commodities; the price change data table is used for storing the types of the fresh commodities, the freshness grades of the commodities and corresponding price values;
the freshness evaluation model is obtained by training a plurality of stacked similar fresh commodity sample images, receives the fresh commodity image of the specified commodity placing area shot by the first shooting device, evaluates the integral freshness grade of the specified fresh commodity, and feeds the evaluation result back to the trend analysis module;
the trend analysis module analyzes the whole freshness grade sequence data of each appointed fresh commodity in a certain time period range, calculates the freshness change trend of each appointed fresh commodity, and generates a replenishment instruction or a price adjustment instruction according to the freshness change trend; the trend analysis module is constructed by adopting a neural network and is obtained by training a fitting curve of the whole freshness grade of the fresh commodity changing along with time under different external factors; the input of the trend analysis module is historical freshness grade data of fresh commodities corresponding to m sampling periods counted from the current moment forward, and the output is estimated freshness grade data of m sampling periods counted from the current moment backward;
and the price adjusting module is combined with the price adjusting instruction and the price change data table to update the price label of the corresponding commodity.
In order to optimize the technical scheme, the specific measures adopted further comprise:
further, the freshness evaluation model comprises a feature extraction unit, a first feature fusion unit, a second feature fusion unit, an adaptive maximum pooling layer, a global average pooling layer and a full connection layer which are connected in sequence;
the feature extraction unit comprises three convolution modules and is used for carrying out feature extraction on the imported fresh commodity image of the appointed goods placement area layer by layer; the first feature fusion unit is used for extracting the outputs of the second convolution module and the third convolution module and carrying out feature fusion of different sizes; the second feature fusion unit is used for performing feature fusion again on the output of the first feature fusion unit and the output of the first convolution module; the characteristics fused by the second characteristic fusion unit are subjected to adaptive maximum pooling and global average pooling respectively through an adaptive maximum pooling layer and a global average pooling layer, and then global characteristics of the fresh commodity image are extracted; the full-connection layer is used for classifying the extracted global features of the fresh commodity images, calculating the probability that the global features of the fresh commodity images belong to each freshness grade, and outputting the freshness grade of the current fresh commodity image.
Further, the acquisition process of the training samples adopted by the freshness evaluation model comprises the following steps:
according to different quantities and different freshness grades of specified fresh commodities, stirring at intervals, and collecting in batches at stirring intervals to obtain initial sample images;
and enhancing the initial sample images of each batch, and expanding the data set by adopting various methods such as scaling, rotating, transforming and turning.
Furthermore, the remote server comprises an electronic tag management module, a commodity management module, a base station management module and a price updating module; the label management module is used for managing the electronic label; the base station management module is used for managing a base station; the commodity management module is used for managing the commodity type, the supply quantity and the selling quantity of the fresh area; the price updating module is used for receiving or generating price labels of all commodities and sending the price labels to the electronic labels through the base station for displaying; when one electronic tag corresponds to a plurality of commodities, the price updating module is also used for splicing the price tags of the commodities and sending the spliced tag images to the electronic tag for display.
Furthermore, the automatic management system also comprises second shooting devices distributed in the fresh area;
the second shooting device is used for shooting shelf images which belong to the unit goods area within a preset visual field range by taking the unit goods area corresponding to each electronic tag as a center;
the price updating module comprises a goods identifying unit, a placing area analyzing unit and a label image splicing unit;
the goods identification unit primarily divides goods shelf images by taking the unit goods areas as units, judges the similarity of the images of the adjacent unit goods areas, regards the adjacent unit goods areas with the similarity higher than a similarity threshold value as the same goods placing area, primarily integrates the unit goods areas on the goods shelf images to generate a plurality of goods placing areas, and identifies each goods placing area; then calling a commodity identification model to identify the commodity type of each goods placing area;
the goods placing area analysis unit defines the goods placing area corresponding to only one electronic tag as a first placing area, defines the rest goods placing areas as second placing areas and determines the main tag of each second placing area according to the image area;
the label image splicing unit is used for calling a commodity label image corresponding to each goods placing area, splicing the called commodity label images according to the position relation of each goods placing area on the goods shelf image, and generating an electronic label price board; and the label images of the second placing area except the main label are zoomed according to the occupied area of the second placing area on the shelf image.
Further, when one goods placing area contains more than two kinds of goods, the goods types contained in the goods placing area are matched with the goods types of the adjacent goods placing area, if all the goods types are successfully matched, the goods placing area is cleared, and otherwise, a corresponding goods placing area is generated according to the goods types which are not successfully matched; the adjacent goods placing areas are the goods placing areas which belong to the same electronic tag display coverage range.
Further, the label image splicing unit comprises a commodity label calling component, a template generating component, a judging component, a zooming component and a filling component;
the commodity label calling assembly reads the goods placing area number and the commodity type on each goods placing area contained in the current acquisition cycle aiming at each electronic label, and calls the commodity label image data corresponding to each goods placing area according to the commodity type;
the template generating assembly generates a label module according to the position information of the unit goods area occupied by each goods placing area, the label template comprises a plurality of label filling frames, and each label filling frame corresponds to one goods placing area;
the judging component is used for judging whether the electronic tag corresponds to a second placing area except the main tag, if so, the zooming component is called to zoom the corresponding tag filling frame in a preset image size range according to the occupied area of the second placing area on the shelf image;
and the filling component fills the called commodity label image data into the corresponding label filling frame to generate the electronic label price tag in the current acquisition period.
Further, the verification module comprises a label image acquisition unit, a verification information calling unit, a first judgment unit, a scaling recalculation unit, a second judgment unit, a third judgment unit and a judgment result output unit;
the label image acquisition unit is combined with the electronic label and the goods placing area relation data table to acquire a display image data set of the electronic label containing the same second placing area a
The verification information calling unit calls the originally acquired goods shelf image data set corresponding to the electronic tag
And according to the position information of the unit goods area occupied by the second placing area a, collecting the display image data of the electronic tag
The image data set { a of the corresponding second putting area a is obtained by extraction(m)}；m＝1，2，...，M；
The first judgment unit judges an image data set { a }(m)Whether the image data of the commodity label in the item is complete or not is judged, if so, a starting signal is sent to a scaling recalculation unit, and if not, a judgment result output unit is called to output an updating failure result;
the scaling recalculation unit calculates the shelf image data set of the second placement area a
The ratio area of each shelf image in the second placing area a is calculated to obtain a scaling set of the label filling frame of the second placing area a
The second judging unit is set according to the scaling ratio
For dataImage data set { a(m)Zooming the image in the verification set to obtain a zoomed verification set { b }(m)}，
Judging the verification set b again(m)Judging whether the sizes of the images in the image group are consistent, if so, sending a starting signal to a third judging unit, and otherwise, calling a judging result output unit to output an updating failure result;
the third judging unit pairs verification set { b }(m)Judging the similarity between the image in the area and the commodity price label of the corresponding commodity in the second placing area a, if the similarity is higher than a preset similarity threshold value, calling a judgment result output unit to output a successful updating result, otherwise, calling the judgment result output unit to output a failed updating result.
The invention has the beneficial effects that:
first, the system for automatically managing the large-screen electronic tag in the fresh area can estimate the freshness grade of the commodities in the fresh area, provide a solution for replenishment or price reduction sales promotion according to the estimation result, automatically adjust the price tag of the commodity according to the sales promotion scheme preset by a user, and greatly reduce the time for manually adjusting and confirming the information of the commodity.
Secondly, the system for automatically managing the electronic tags on the large screen of the fresh area can automatically adjust the display images of the electronic tags on the large screen according to the real-time state of the commodities on the goods shelf; by taking the visual range of the user as a reference, the user can acquire the price information of the commodity type (including misplaced commodities) in the close visual range at any angle, and the user experience is improved. In addition, because the price label display content can contain the price information of all misplaced commodities, the probability of manual rearrangement can be reduced in a mode of attracting users to purchase.
Thirdly, the system for automatically managing the electronic tags on the large screen of the fresh area can automatically verify the updating result by combining the repeated price tags of part of commodities, and the reliability of the verification result is high.
Drawings
Fig. 1 is a structural diagram of an automatic management system for a large-screen electronic tag of a fresh food area according to the invention.
FIG. 2 is a diagram of a freshness evaluation model of the present invention.
Fig. 3 is a block diagram of a price update module of the present invention.
FIG. 4 is a block diagram of an authentication module of the present invention.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
It should be noted that the terms "upper", "lower", "left", "right", "front", "back", etc. used in the present invention are for clarity of description only, and are not intended to limit the scope of the present invention, and the relative relationship between the terms and the terms is not limited by the technical contents of the essential changes.
Fig. 1 is a structural diagram of an automatic management system for a large-screen electronic tag of a fresh food area according to the invention. Referring to fig. 1, the automatic management system is used for managing and updating display contents of electronic tags distributed and installed in a fresh food area, and comprises a base station, a first shooting device, an area management and control device and a remote server.
The remote server locally generates a price tag of each fresh area commodity and sends the price tag to a corresponding electronic tag through the base station; the electronic tag adopts an electronic paper display screen and is used for displaying the price tag of at least one fresh area commodity on the shelf.
First shooting device installs in electronic tags below, links to each other with regional management and control device, and first shooting device shoots the bright commodity image of giving birth to in appointed goods that electronic tags corresponds and puts the district, sends the shooting result to regional management and control device.
The region control device comprises a price change data table, a price adjusting module, a trend analyzing module and a plurality of freshness evaluation models of fresh commodities; the price change data table is used for storing the types of the fresh commodities, the freshness grades of the commodities and corresponding price values.
The fresh degree evaluation model adopts a plurality of same-class fresh commodity sample images in a stacked form to be trained to obtain, receives the fresh commodity image in the appointed goods placing area shot by the first shooting device, evaluates the whole fresh degree grade of the appointed fresh commodity, and feeds the evaluation result back to the trend analysis module.
The trend analysis module analyzes the whole freshness grade sequence data of each appointed fresh commodity in a certain time period range, calculates to obtain the freshness change trend of the appointed fresh commodity, and generates a replenishment instruction or a price adjustment instruction according to the freshness change trend; the trend analysis module is obtained by training the whole freshness grade sequence data and the sales sequence data which take unit time as a reference.
The price adjusting module is combined with the price adjusting instruction and the price change data table to update the price label of the corresponding commodity.
First, freshness judgment
Referring to fig. 2, the freshness evaluation model includes a feature extraction unit, a first feature fusion unit, a second feature fusion unit, an adaptive maximum pooling layer, a global average pooling layer, and a full connection layer, which are connected in sequence.
The feature extraction unit comprises three convolution modules and is used for carrying out feature extraction on the imported fresh commodity image of the appointed goods placement area layer by layer; the first feature fusion unit is used for extracting the outputs of the second convolution module and the third convolution module and carrying out feature fusion of different sizes; the second feature fusion unit is used for performing feature fusion again on the output of the first feature fusion unit and the output of the first convolution module; the characteristics fused by the second characteristic fusion unit are subjected to adaptive maximum pooling and global average pooling respectively through an adaptive maximum pooling layer and a global average pooling layer, and then global characteristics of the fresh commodity image are extracted; the full connecting layer is used for carrying out classification processing on the extracted global features of the fresh commodity images, calculating the probability that the global features of the fresh commodity images belong to each freshness grade, and outputting the freshness grade of the current fresh commodity images.
The first feature fusion unit is used for fusing the middle layer feature value output by the second convolution module and the deep layer feature value output by the third convolution module to obtain a first fusion feature used for representing semantic information of the fresh commodity. The second feature fusion unit is used for fusing the middle layer feature value output by the second convolution module and the deep layer feature value output by the third convolution module to obtain a first fusion feature. The first convolution module focuses more on extracting geometric detail features of the fresh commodity. And performing secondary feature fusion on the first fusion feature (more emphasizing on the epidermis semantic feature of the fresh commodity) and the shallow feature value (more emphasizing on the geometric detail feature of the fresh commodity) output by the first convolution module by adopting a second feature fusion unit to obtain the global feature of the image of the fresh commodity. In addition, the embodiment judges the whole freshness of the fresh commodity, simultaneously considers the accuracy and the efficiency of the evaluation result, and can be executed only by adopting three layers of convolution networks, so that the lightweight operation requirement of the fresh area is met.
The acquisition process of the training samples adopted by the freshness evaluation model comprises the following steps:
according to different quantities and different freshness grades of specified fresh commodities, stirring at intervals, and collecting in batches at stirring intervals to obtain initial sample images; and enhancing the initial sample images of each batch, and expanding the data set by adopting various methods such as scaling, rotating, transforming and turning. For example, the freshness degree level of a certain fresh commodity is divided into five levels of 100%, 80%, 60%, 40% and 20%, four numbers of 200, 100, 50 and 20 are respectively set for the five freshness degree levels, and 20 samples of the original fresh commodity are summed up. And (3) stirring the 20 original fresh commodity samples, carrying out image acquisition in a stirring gap, and circulating the steps to acquire initial sample image data. After the initial sample image data is enhanced, the original sample image data is expanded by scaling, rotating, transforming and turning, and 16400 fresh image sample data with different resolutions can be finally obtained. In practical applications, the freshness degree grade and the number grade can be set by themselves, and are not limited to the above.
In view of the judgment efficiency, the present embodiment proposes to perform the freshness judgment on the whole of the fresh goods, rather than on the individual fresh goods. In the business surpassing, environmental factors such as light are kept consistent, even if partial commodity environmental light is reduced due to reasons such as human body shielding, the influence of light intensity on freshness judgment can be ignored only by correspondingly screening collected images. On this basis, the freshness change of fresh goods is related to two factors: the loss of water diversion over time, customer selection, results in a change in the freshness portion ratio of the remaining commodity. In the former case, the curve of the change in freshness due to water loss is relatively flat, and in the latter case, the customer is more likely to select a fresh product, and therefore, the behavior of purchasing the product by the customer causes the change in freshness of the remaining product. For example, for a certain fresh commodity, such as apple, since the commercial warehouse is usually provided with a fresh-keeping area, we default that the apple just put in the goods placement area is fresh (judged from an overall perspective, not a single apple), as time goes by, moisture of the individual apple is lost, and as customers are more inclined to select fruits with higher freshness, the overall freshness grade of the apple in the goods placement area is continuously reduced.
On the basis of obtaining the real-time freshness of the fresh goods, the price label adjustment or the replenishment request can be executed according to the real-time freshness, but more preferably, the freshness of the fresh goods is estimated according to the historical freshness grade data, and the follow-up action is executed according to the estimation result. The embodiment adopts a GRU network model based on an attention mechanism to achieve the technical effect. The input of the trend analysis module is historical freshness grade data of fresh commodities corresponding to m sampling periods counted from the current moment forward, and the output is estimated freshness grade data of m sampling periods counted from the current moment backward. Preferably, the trend analysis module is trained by using the freshness grade sequence data of the fresh goods in a certain time period range and the freshness change sample data caused by different sales volumes under various freshness grade conditions.
The trend analysis model of the present embodiment is based on the following shopping assumptions from historical data analysis: the sales volume at the same freshness grade tends to be stable, but is related to external factors such as season, price level and the number of like commodities provided by the current supermarket. When the overall freshness grade of the fresh goods is high, the shopping desire of a user can be excited, the sales volume in unit time is high, and the change curve of the overall freshness grade along with time contains frequent mutation; along with the lapse of time and the selling of commodity, the whole new freshness grade of giving birth to bright commodity reduces gradually, and user's shopping desire also more and more hangs down, and unit interval sales volume also can walk down gradually, contains less sudden change in the change curve of whole new freshness grade along with time, and the curve can be gentler. Therefore, when external factors are determined, the change curve of the overall freshness grade of the fresh goods is relatively fixed. In the embodiment, historical freshness grade data of fresh goods corresponding to m sampling periods are collected, and meanwhile estimated freshness grade data of m sampling periods counted backwards from the current moment are estimated.
Illustratively, the trend analysis module can be constructed by adopting a neural network and is obtained by training a fitting curve of the overall freshness grade of the fresh commodity changing along with time under different external factors.
Still taking apple as an example, the training process of the trend analysis module is as follows: under the condition of different external factors, acquiring the whole freshness grade sequence data of the apples within a certain time length range, fitting to obtain a whole freshness grade curve of the apples within the time length range as a group of sample data by taking time as a horizontal axis and taking the whole freshness grade as a longitudinal axis. And training the trend analysis module by adopting the sample data set, so that the trend analysis module can predict the freshness change curve of the apples in the next m sampling periods according to the historical freshness grade data of the apples in the imported m sampling periods, and executing price label adjustment or generating a replenishment request according to the prediction result. For example, when the freshness level of a commodity is reduced to 40% freshness level, the discount price corresponding to 40% of the freshness level in the price change data table is consulted, and the price of the electronic tag of the commodity is automatically updated; or when the freshness grade of the commodities in a short period is estimated to be rapidly reduced to 40% freshness grade, the commodities are replenished, and the phenomena of selling neutral gear and missing gold selling time are avoided.
Preferably, the embodiment further provides a method for managing commodity prices with the maximized sales as an optimization target, specifically, a change curve of the overall freshness grade and the sales volume of fresh commodities under different external factor conditions is obtained by combining historical data, an estimated sales volume change curve of the commodities is obtained by analyzing the estimated freshness grade change curve, and a corresponding price adjustment strategy is automatically executed according to an estimated result. For example, when the freshness degree of a product is decreased to 40% and the freshness is maintained slowly in an estimated time period, which indicates that the sales volume is difficult to be promoted in a short time period, the price of the electronic tag of the product is automatically updated by referring to the discount price corresponding to 40% of the freshness degree in the price change data table. Conversely, if the freshness level of the product is reduced to 40% but the freshness level is also reduced to a greater extent within the estimated time period, which indicates that even if the freshness is low, the estimated sales volume of the product is still greater and the customer's psychological acceptance of the product is still higher, then the customer may choose to continue to look at the product. The embodiment can reduce manual intervention, reversely deduces the commodity sales volume by estimating the freshness grade, manages the commodity price by taking the maximum sales volume as an optimization target, and reduces the workload of manually checking commodities and manually updating price labels (even updating electronic labels). The price change data table may be maintained updated by a remote server.
Two, multiple commodity price label display
The fresh area can adopt the goods platform structure of medium height and can set up tens or even tens unit goods district on a goods platform because goods are of a great variety usually. The merchant sets up an electronic tag about 10 inches for each commodity, and places the electronic tag right in front of the corresponding commodity due to price indication. The user needs to check the price of the commodity after the commodity corresponds to the electronic tag one by one. For this purpose, the present embodiment proposes that a plurality of item prices are displayed simultaneously on one electronic label.
In this embodiment, the display content of the electronic tag is affected by the visual range of the user, that is, the electronic tag in each fresh area is to be able to display the prices of other commodities within a certain visual range of the user with the corresponding commodity as the center, and compared with the electronic tags which are set in one-to-one correspondence with the quantity of the commodities according to the commodity placement positions, the effort of the user to find the prices is greatly reduced.
The collection of goods shelves image can be through distributing the second shooting device realization in giving birth to the bright district, also can combine artifical tour to shoot in step, if needs combine artifical tour to shoot, the image of artifical tour shooting can carry out the school again, or indirect calculation its and fixed point position shoot the correlation between the image, use again as price updating application. In this embodiment, the price updating may be performed when the goods are put on shelf, and may also be performed in a super-commercial routine operation and maintenance, for example, when one of the goods on a certain shelf is sold out, or the price tag may be adjusted in time by periodic updating to avoid misunderstanding of the user.
Referring to fig. 3, the price updating module includes a goods identifying unit, a putting area analyzing unit, and a label image splicing unit.
The goods identification unit primarily divides goods shelf images by taking the unit goods areas as units, judges the similarity of the images of the adjacent unit goods areas, regards the adjacent unit goods areas with the similarity higher than a similarity threshold value as the same goods placing area, primarily integrates the unit goods areas on the goods shelf images to generate a plurality of goods placing areas, and identifies each goods placing area; and then calling a commodity identification model to identify the commodity type of each commodity placing area.
The process of identifying the goods information by the goods identification unit comprises the following substeps:
and collecting a plurality of commodity image samples contained in the commodity excessiveness, and constructing a training set, wherein the commodity image samples contain stacked image samples of a plurality of similar commodities. And constructing a commodity identification model based on the neural network, and training the commodity identification model by adopting a training set. The commodity identification model may use an image identification technology in the prior art, for example, the invention disclosed in application No. 202010874612.9 published on 10/02/2020 discloses a commodity identification management method, and the commodity type and model information is obtained by training and identifying an image of a target commodity. Primarily dividing goods shelf images by taking the unit goods areas as units, judging the similarity of the images of the adjacent unit goods areas, and regarding the adjacent unit goods areas with the similarity higher than a similarity threshold value as the same goods placement area; and preliminarily integrating unit goods areas on the goods shelf image, generating a plurality of goods placing areas, and identifying each goods placing area. And (4) importing the image of each goods placing area into a goods identification model, and identifying the goods type of each goods placing area.
When one goods placing area contains more than two kinds of goods, the goods types contained in the goods placing area are matched with the goods types of the adjacent goods placing areas, if all the goods types are successfully matched, the goods placing area is cleared, and otherwise, a corresponding goods placing area is generated according to the goods types which are not successfully matched; the adjacent goods placing areas are the goods placing areas which belong to the same electronic tag display coverage range. This is because, in the shopping environment, the user often takes the product and misplaces the product in another area because of the open shopping environment. In the prior art, for this situation, usually, a misplaced goods is found through manual inspection or image recognition combined with database proofreading, and then a worker is notified to correct the position of the goods in a manner of putting the goods again. In the embodiment, the commodity is not regarded as a pure misplaced commodity, but is regarded as a secondary selected commodity, and the price information of the secondary selected commodity is displayed in the large-screen electronic tag, so that the probability of purchasing the commodity by the user is increased, and the picking and placing workload of workers is reduced.
The goods placing area analysis unit defines the goods placing area corresponding to one electronic tag as a first placing area, defines the rest goods placing areas as second placing areas and determines the main tag of each second placing area according to the image area.
In this embodiment, the commodities in the second placing area include commodities which are not easy to attract the attention of the user due to a small occupied area in the same viewing angle direction, in addition to the secondary selected commodities which are misplaced. The product occupies a small area and can only be partially displayed due to an angle problem in the viewing angle direction a, but it is likely to be completely displayed in the viewing angle direction B (for example, when an electronic tag is located right in front of the product 1, and therefore, when a shelf image is collected with the orientation of the electronic tag as the viewing angle center, the product 1 occupies a large area, so that the electronic tag becomes a main tag of the product 1). Of course, in some examples, the main label of the product may also be specified by the relationship between the background binding label and the product; however, the confirmation process of the former main label is more consistent with the visual law of the user during shopping, and particularly, when the commodity is displaced, and a plurality of unit goods areas are repeatedly placed in the commodity or the label position is deviated, the self-correction of the price display of the electronic label can still be realized. The characteristics of multi-commodity display are combined, and on the basis of ensuring that commodity price display is complete, a client can be ensured to continuously and quickly obtain correct price information.
The label image splicing unit is used for calling a commodity label image corresponding to each goods placing area, splicing the called commodity label images according to the position relation of each goods placing area on the goods shelf image, and generating an electronic label price board; and the label images of the second placing area except the main label are zoomed according to the occupied area of the second placing area on the shelf image.
The label image splicing unit comprises a commodity label calling component, a template generating component, a judging component, a zooming component and a filling component. The goods label calling assembly reads the goods placing area number and the goods type on each goods placing area contained in the current acquisition cycle aiming at each electronic label, and calls the goods label image data corresponding to each goods placing area according to the goods type; the template generating component generates a label module according to the position information of the unit goods area occupied by each goods placing area, the label template comprises a plurality of label filling frames, and each label filling frame corresponds to one goods placing area; the judging component is used for judging whether the electronic tag corresponds to a second placing area except the main tag or not, and if so, calling the zooming component to zoom the corresponding tag filling frame in a preset image size range according to the occupied area of the second placing area on the shelf image; and the filling component fills the called commodity label image data into the corresponding label filling frame to generate the electronic label price tag in the current acquisition period.
After the corresponding relation between the electronic tag and the goods placing area is obtained, the splicing process of the commodity price tag can be realized according to the position relation of the goods placing area. The electronic tag price board is formed by splicing according to the position relation of commodities, and can correctly guide customers to know commodity information; meanwhile, the commodity label image in the electronic label price tag is related to the attention of the user under the current visual angle condition, so that the energy required for the user to obtain price information can be reduced, and the misjudgment rate is reduced.
In the embodiment, although the commodity label image is zoomed, the image is still zoomed within a reasonable range of human vision, and the size and the number of the electronic labels corresponding to the shelf can be determined according to the commodity type and the number of the shelf.
And finally, the label image splicing unit sends the spliced electronic label price tags to corresponding electronic labels for display.
In this embodiment, the purpose of setting the second display area is to enable the user to completely and quickly know the prices of all the commodities in the visual range, and the other important meaning is the verification standard as the updating result.
At present, the verification of the display result of the price label is still realized manually, the types of commodities in the fresh area are numerous, the probability of sales promotion and discount is high, and even the staff with rich experience is difficult to remember the real-time prices of most commodities. If the updates are frequent, the workload of the worker will increase geometrically. This is especially true for large-screen price labels of multi-item price displays. Therefore, the embodiment proposes that consistency verification is performed on repeatedly displayed price label image data of the second placing area, and estimation of the whole electronic label display result is achieved.
Referring to fig. 4, the remote server further comprises a verification module; the verification module comprises a label image acquisition unit, a verification information calling unit, a first judgment unit, a scaling recalculation unit, a second judgment unit, a third judgment unit and a judgment result output unit;
the label image acquisition unit is combined with the electronic label and the goods placing area relation data table to acquire a display image data set of the electronic label comprising the same second placing area a
The collecting method comprises the steps of directly shooting the price board image of the electronic tag or requesting the electronic tag to feed back price board image data through the base station.
The verification information calling unit calls the originally acquired goods shelf image data set corresponding to the electronic tag
And according to the position information of the unit goods area occupied by the second placing area a, collecting the display image data of the electronic tag
Extracting the image data set { a of the corresponding second placing area a(m)}；m＝1，2，...，M。
The first judging unit judges the image data set { a }(m)And (4) judging whether the image data of the commodity label in the item is complete, if so, sending a starting signal to a scaling recalculation unit, and otherwise, calling a judgment result output unit to output an update failure result.
The scaling recalculation unit calculates the shelf image data set of the second placing area a
The ratio area of each shelf image in the second placing area a is calculated to obtain a scaling set of the label filling frame of the second placing area a
The second judging unit is set according to the scaling ratio
For a data image data set { a(m)Zooming the image in the verification set to obtain a zoomed verification set { b }(m)}，
Judge the verification set { b(m)And judging whether the image sizes are consistent, if so, sending a starting signal to a third judging unit, and otherwise, calling a judging result output unit to output an updating failure result.
Third judging unit pair verification set { b(m)Judging the similarity between the image in the area and the commodity price label of the corresponding commodity in the second placing area a, if the similarity is higher than a preset similarity threshold value, calling a judgment result output unit to output a successful updating result, otherwise, calling the judgment result output unit to output a failed updating result.
Taking the four-side cargo bed as an example, assuming that a plurality of commodities 1 are placed on the cargo bed located on the east side, partial commodities 1 can be observed from the south and north aisles, the four-side cargo bed comprises three electronic tags, wherein one electronic tag is opposite to the east side and corresponds to the commodity 1, and the other two electronic tags face the southwest side and the northwest side respectively and correspond to the commodities 2 and 3. Through analysis, the electronic tag 1 facing east is a main tag, and the electronic tags 2 and 3 facing southwest and northwest are auxiliary tags. The electronic tags 1, 2, and 3 process and display the product price image of the product 1 according to the area and position of the product 1 on the shelf image corresponding thereto. In the electronic tag 2, the price image of the product 1 is displayed to the right with a small size, and in the electronic tag 3, the price image of the product 1 is displayed to the left with a small size.
In the verification process, complete display images of the electronic tags 1, 2 and 3 are respectively collected, corresponding price images of the placing areas are intercepted from the set positions in the complete display images according to the electronic tag templates, if the price images of the placing areas are incomplete, the updating of the current electronic tags fails, and the updated image data is probably not received; otherwise, the original shelf image is processed again, the scaling is calculated, the intercepted price image of the placement area is scaled and then compared with the consistency, and meanwhile, the calculation result, the scaling and the image data display of the original shelf image are verified. The original shelf image is a data basis when the electronic tag is updated, and not only an image when the commodity is put on the shelf, which is for the purpose of verification work when the electronic tag is updated in the commodity selling process.
The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above-mentioned embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may be made by those skilled in the art without departing from the principle of the invention.
Claims (8)
1. An automatic management system for a large-screen electronic tag in a fresh area is used for managing and updating display contents of electronic tags distributed and installed in the fresh area; the automatic management system is characterized by comprising a base station, a first shooting device, an area management and control device and a remote server;
the remote server locally generates a price tag of each fresh area commodity and sends the price tag to a corresponding electronic tag through the base station; the electronic tag adopts an electronic paper display screen and is used for displaying the price tag of at least one fresh area commodity on the shelf;
the first shooting device is arranged below the electronic tag and connected with the area control device, shoots fresh commodity images of the specified goods placing area corresponding to the electronic tag and sends shooting results to the area control device;
the region control device comprises a price change data table, a price adjusting module, a trend analyzing module and a plurality of freshness evaluation models of fresh commodities; the price change data table is used for storing the types of the fresh commodities, the freshness grades of the commodities and corresponding price values;
the freshness evaluation model is obtained by training a plurality of stacked similar fresh commodity sample images, receives the fresh commodity image of the specified commodity placing area shot by the first shooting device, evaluates the integral freshness grade of the specified fresh commodity, and feeds the evaluation result back to the trend analysis module;
the trend analysis module analyzes the whole freshness grade sequence data of each appointed fresh commodity in a certain time period range, calculates the freshness change trend of each appointed fresh commodity, and generates a replenishment instruction or a price adjustment instruction according to the freshness change trend; the trend analysis module is constructed by adopting a neural network and is obtained by training a fitting curve of the whole freshness grade of the fresh commodity changing along with time under different external factors; the input of the trend analysis module is historical freshness grade data of fresh commodities corresponding to m sampling periods counted from the current moment forward, and the output is estimated freshness grade data of m sampling periods counted from the current moment backward;
and the price adjusting module is combined with the price adjusting instruction and the price change data table to update the price label of the corresponding commodity.
2. The fresh area large-screen electronic tag automatic management system according to claim 1, wherein the freshness evaluation model comprises a feature extraction unit, a first feature fusion unit, a second feature fusion unit, an adaptive maximum pooling layer, a global average pooling layer and a full connection layer which are connected in sequence;
the feature extraction unit comprises three convolution modules and is used for carrying out feature extraction on the imported fresh commodity image of the appointed goods placement area layer by layer; the first feature fusion unit is used for extracting the outputs of the second convolution module and the third convolution module and carrying out feature fusion of different sizes; the second feature fusion unit is used for performing feature fusion again on the output of the first feature fusion unit and the output of the first convolution module; the characteristics fused by the second characteristic fusion unit are subjected to adaptive maximum pooling and global average pooling respectively through an adaptive maximum pooling layer and a global average pooling layer, and then global characteristics of the fresh commodity image are extracted; the full-connection layer is used for classifying the extracted global features of the fresh commodity images, calculating the probability that the global features of the fresh commodity images belong to each freshness grade, and outputting the freshness grade of the current fresh commodity image.
3. The fresh area large screen electronic tag automatic management system according to claim 1, wherein the collection process of the training samples adopted by the freshness evaluation model comprises:
according to different quantities and different freshness grades of specified fresh commodities, stirring at intervals, and collecting in batches at stirring intervals to obtain initial sample images;
and enhancing the initial sample images of each batch, and expanding the data set by adopting various methods such as scaling, rotating, transforming and turning.
4. The automatic management system for the large-screen electronic tags in the fresh food areas of any one of claims 1 to 3, wherein the remote server comprises an electronic tag management module, a commodity management module, a base station management module and a price updating module; the label management module is used for managing the electronic label; the base station management module is used for managing a base station; the commodity management module is used for managing the commodity type, the supply quantity and the selling quantity of the fresh area; the price updating module is used for receiving or generating price labels of all commodities and sending the price labels to the electronic labels through the base station for displaying; when one electronic tag corresponds to a plurality of commodities, the price updating module is also used for splicing the price tags of the commodities and sending the spliced tag images to the electronic tag for display.
5. The large-screen electronic tag automatic management system for fresh food areas of claim 4, further comprising second photographing devices distributed in the fresh food areas;
the second shooting device is used for shooting shelf images which belong to the unit goods area within a preset visual field range by taking the unit goods area corresponding to each electronic tag as a center;
the price updating module comprises a goods identifying unit, a placing area analyzing unit and a label image splicing unit;
the goods identification unit primarily divides goods shelf images by taking the unit goods areas as units, judges the similarity of the images of the adjacent unit goods areas, regards the adjacent unit goods areas with the similarity higher than a similarity threshold value as the same goods placing area, primarily integrates the unit goods areas on the goods shelf images to generate a plurality of goods placing areas, and identifies each goods placing area; then calling a commodity identification model to identify the commodity type of each goods placing area;
the goods placing area analysis unit defines the goods placing area corresponding to only one electronic tag as a first placing area, defines the rest goods placing areas as second placing areas and determines the main tag of each second placing area according to the image area;
the label image splicing unit is used for calling a commodity label image corresponding to each goods placing area, splicing the called commodity label images according to the position relation of each goods placing area on the goods shelf image, and generating an electronic label price board; and the label images of the second placing area except the main label are zoomed according to the occupied area of the second placing area on the shelf image.
6. The automatic management system for the large-screen electronic tags in the fresh food area as claimed in claim 5, wherein when one of the goods placing areas contains more than two kinds of goods, the goods types contained in the goods placing area are matched with the goods types of the adjacent goods placing areas, if all the goods types are successfully matched, the goods placing area is cleared, otherwise, a corresponding goods placing area is generated according to the goods types which are not successfully matched; the adjacent goods placing areas are the goods placing areas which belong to the same electronic tag display coverage range.
7. The fresh area large-screen electronic tag automatic management system according to claim 5, wherein the tag image stitching unit comprises a commodity tag calling component, a template generating component, a judging component, a zooming component and a filling component;
the commodity label calling assembly reads the goods placing area number and the commodity type on each goods placing area contained in the current acquisition cycle aiming at each electronic label, and calls the commodity label image data corresponding to each goods placing area according to the commodity type;
the template generating assembly generates a label module according to the position information of the unit goods area occupied by each goods placing area, the label template comprises a plurality of label filling frames, and each label filling frame corresponds to one goods placing area;
the judging component is used for judging whether the electronic tag corresponds to a second placing area except the main tag, if so, the zooming component is called to zoom the corresponding tag filling frame in a preset image size range according to the occupied area of the second placing area on the shelf image;
and the filling component fills the called commodity label image data into the corresponding label filling frame to generate the electronic label price tag in the current acquisition period.
8. The large-screen electronic tag automatic management system for fresh food areas of claim 5, wherein the remote server further comprises a verification module; the verification module comprises a label image acquisition unit, a verification information calling unit, a first judgment unit, a scaling recalculation unit, a second judgment unit, a third judgment unit and a judgment result output unit;
the label image acquisition unit is combined with the electronic label and the goods placing area relation data table to acquire a display image data set of the electronic label containing the same second placing area a
The verification information calling unit calls the number of the originally acquired shelf images corresponding to the electronic tagAccording to a set
And according to the position information of the unit goods area occupied by the second placing area a, collecting the display image data of the electronic tag
Extracting the image data set { a of the corresponding second placing area a(m)}；m＝1，2，...，M；
The first judgment unit judges an image data set { a }(m)Whether the image data of the commodity label in the item is complete or not is judged, if so, a starting signal is sent to a scaling recalculation unit, and if not, a judgment result output unit is called to output an updating failure result;
the scaling recalculation unit calculates the shelf image data set of the second placement area a
The ratio area of each shelf image in the second placing area a is calculated to obtain a scaling set of the label filling frame of the second placing area a
The second judging unit is set according to the scaling ratio
For a data image data set { a(m)Zooming the image in the verification set to obtain a zoomed verification set { b }(m)}，
Judging the verification set b again(m)Judging whether the sizes of the images in the image group are consistent, if so, sending a starting signal to a third judging unit, and otherwise, calling a judging result output unit to output an updating failure result;
the third judging unit checksCertificate set { b(m)Judging the similarity between the image in the area and the commodity price label of the corresponding commodity in the second placing area a, if the similarity is higher than a preset similarity threshold value, calling a judgment result output unit to output a successful updating result, otherwise, calling the judgment result output unit to output a failed updating result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111607248.0A CN114282816A (en) | 2021-12-24 | 2021-12-24 | Give birth to bright district large screen electronic tags automatic management system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111607248.0A CN114282816A (en) | 2021-12-24 | 2021-12-24 | Give birth to bright district large screen electronic tags automatic management system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114282816A true CN114282816A (en) | 2022-04-05 |
Family
ID=80875674
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111607248.0A Pending CN114282816A (en) | 2021-12-24 | 2021-12-24 | Give birth to bright district large screen electronic tags automatic management system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114282816A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114786172A (en) * | 2022-04-14 | 2022-07-22 | 广州计量检测技术研究院 | Automatic upgrading method and system for electronic tag and electronic tag |
| CN116308473A (en) * | 2022-12-30 | 2023-06-23 | 佛山市金仓联货架制造有限公司 | Automatic price adjustment method and automatic price adjustment device for electronic price tag of goods shelf |
| CN117152539A (en) * | 2023-10-27 | 2023-12-01 | 浙江由由科技有限公司 | Fresh commodity classification correction method based on dimension reduction feature machine verification |
-
2021
- 2021-12-24 CN CN202111607248.0A patent/CN114282816A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114786172A (en) * | 2022-04-14 | 2022-07-22 | 广州计量检测技术研究院 | Automatic upgrading method and system for electronic tag and electronic tag |
| CN116308473A (en) * | 2022-12-30 | 2023-06-23 | 佛山市金仓联货架制造有限公司 | Automatic price adjustment method and automatic price adjustment device for electronic price tag of goods shelf |
| CN117152539A (en) * | 2023-10-27 | 2023-12-01 | 浙江由由科技有限公司 | Fresh commodity classification correction method based on dimension reduction feature machine verification |
| CN117152539B (en) * | 2023-10-27 | 2024-01-26 | 浙江由由科技有限公司 | Fresh commodity classification correction method based on dimension reduction feature machine verification |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114282816A (en) | Give birth to bright district large screen electronic tags automatic management system | |
| US11151427B2 (en) | Method and apparatus for checkout based on image identification technique of convolutional neural network | |
| CN108320404B (en) | Commodity identification method and device based on neural network and self-service cash register | |
| CN108537166B (en) | Method and device for determining shelf browsing amount and analyzing browsing amount | |
| WO2020010821A1 (en) | Method and device for restocking management for automatic vending machine, and user terminal | |
| CN109726759B (en) | Unmanned vending method, device, system, electronic equipment and computer readable medium | |
| CN111222389B (en) | Analysis method and system for commodity on super commodity shelf | |
| CN111310706B (en) | Commodity price tag identification method and device, electronic equipment and storage medium | |
| CN110119914B (en) | Object replenishment processing method, device and system | |
| KR20200119385A (en) | Artificial intelligence appraisal system, artificial intelligence appraisal method and storage medium | |
| CN112508132B (en) | Training method and device for identifying SKU | |
| US10891561B2 (en) | Image processing for item recognition | |
| WO2021179138A1 (en) | Method and system for analyzing commodities on supermarket shelf | |
| CN110428300A (en) | Price mark image processing method, sale and the allocator of fresh commodity | |
| CN113947425A (en) | Shop address determining method and device | |
| CN110826481A (en) | Data processing method, commodity identification method, server and storage medium | |
| CN114282818A (en) | Bright area electronic tag control method for maximizing sales | |
| CN111428743A (en) | Commodity identification method, commodity processing device and electronic equipment | |
| CN117078360A (en) | Intelligent commodity display method, system and device based on off-line popularization | |
| CN109858448A (en) | Item identification method and equipment under a kind of public safety | |
| CN111783627B (en) | Commodity stock determining method, device and equipment | |
| CN112270562B (en) | Automatic price adjustment settlement method and system for agricultural products with commercial and super agricultural products | |
| CN114821729B (en) | Commodity shopping guide method and device, cloud server and storage medium | |
| CN116935198B (en) | Method and device for detecting shortage of goods on goods shelf and robot | |
| CN114185912A (en) | Automatic updating method for commercial and super-large screen electronic tag with verification function |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |